Previous research on predictive factors of marital satisfaction has been limited in several respects. Past work has tended to focus on a population sample of spouses rather than comparative clinical and nonclinical samples, which are obviously quite different groups; in addition, past research has concentrated on examining either personality variables in each spouse, or on each spouse's behaviors during discussion interactions. The main hypotheses I will test in this proposal are intended to address these limitations. To this end, I will propose two models examining the direct and indirect effects of personality variables on marital satisfaction as mediated by social support behaviors, and determine the possible direct and indirect benefits of marital therapy with distressed couples on social support and marital satisfaction. For example, I will examine an individual's personality and use of social support behaviors in predicting their spouse's marital satisfaction, rather than their own, which has never before been studied. Distressed and nondistressed couples will provide self-report data on personality and attachment style, they will also be observed in several dyadic discussions of marital and personal issues. Structural equation models, which are maximally useful in testing simultaneous, mediational, and predictive hypotheses using latent variables, will constitute the basis for data, analysis. This study will not only add to the dearth of research in this area, it may also provide useful information on improving current methods of marital therapy.